Title :
Directional multi-pass median filter for impulse noise reduction
Author :
Vajpai, Jayashri ; Mehta, Sharad
Author_Institution :
M.B.M. Eng. Coll., JNV Univ., Jodhpur, India
Abstract :
A novel two-stage directional multi-pass median filter (DMPMF) for reducing impulse noise has been proposed in this paper. It is a progressive technique wherein for each pass, median is substituted for each noisy pixel even if one noise free pixel exists in its surrounding 3×3 window otherwise the pixel is kept intact. The initial pass replaces only a fraction of the total number of noise-corrupted pixels in the corrupted image. With increasing number of passes, the number of pixels covered increases, and after a few passes, all or most of the noisy pixels are covered. Each pass gradually improves the quality of the restored image. Thereafter, directional filtering is applied to restore the fine details of the final image. The efficacy of the proposed algorithm is determined by applying it to different images. It produces better results in terms of the qualitative and quantitative measures of the image for both low and high density noise in comparison to many existing techniques.
Keywords :
image restoration; impulse noise; median filters; DMPMF; directional multipass median filter; high density noise; image corruption; image restoration; impulse noise reduction; noise-corrupted pixels; Filtering; Filtering algorithms; Image restoration; Noise level; Noise measurement; PSNR; Impulse noise; median filter; multi-pass; noise detection;
Conference_Titel :
Computer Vision, Pattern Recognition, Image Processing and Graphics (NCVPRIPG), 2013 Fourth National Conference on
Conference_Location :
Jodhpur
Print_ISBN :
978-1-4799-1586-6
DOI :
10.1109/NCVPRIPG.2013.6776163